45 
STUDY ON NATIVE XML DATABASE BASED GML STORAGE MODEL 
Shuliang Zhang 3- *, Jiayan Gan 3 , Jiehui Xu 3 ,Guonian Lv a 
a Key lab of Virtual geographic environment, Ministry of education, Nanjing Normal University ,210046,Nanjing, 
China - zhangshuliang@njnu.edu.cn,ganjiayan@126.com 
KEY WORDS: GML, XML, Native XML Database, GML Storage, JTS, XQuery 
ABSTRACT: 
As a mature spatial interoperability specification of OGC, GML rapidly plays an important role in a large number of GIS research 
and application domains, such as spatial data modelling, transmission and exchanging, integration and sharing, because of its 
advantages of opening and self-description format system, rich spatial data expression technology, flexible application schema and 
so on. However, the complexity of GML specification leads to bulkiness of GML data files, which limits extent of GML data sharing. 
Furthermore, its non-structural data model goes against its storage by commercial relational database, which ultimately result in low 
performance of GML application and destroy abundant semantic of GML data. Thus, appropriate GML storage approach would 
improve extend and depth of GML application.By analogy approach, this paper study on storage and manage model of XML data, 
analyze current status of GML storage method. Combining XML database with characteristic of GML spatial data, create GML 
storage mapping mechanism based on feature and storage granularity, form GML to Native XML Database Storage Model 
(G2NXDBM) consisted of GML file segment, GML schema file and GML storage log file. On the basics of JAXP (Java API for 
XML Processing) program API and open source code JTS (Java Topology Suite), the author developed G2NXDBMs prototype 
system which include schema mapping constructor and file storage tools depending on Ipedo native database by IPedo Inc.USA. To 
test the efficiency of GML query, the author selects 5 groups of test data with the same size. The experiment result demonstrates that 
although the storage space of G2NXDBMs is larger than GML file, it has more advantages of query efficiency, support of XML 
technology and data management than GML file and relational database storage, it also testify the validity and practicability of 
G2NXDBM. 
1. INTRODUCTION 
As a mature spatial interoperability specification of OGC, GML 
rapidly plays an important role in a large number of GIS 
research and application domains, such as spatial data 
modelling, transmission and exchanging, integration and 
sharing, because of its advantages of opening and self 
description format system, rich spatial data expression 
technology, flexible application schema and so on. However, 
the complexity of GML specification leads to bulkiness of 
GML data files, which limits extent of GML data sharing. 
Furthermore, its non-structural data model goes against its 
storage by commercial relational database, which ultimately 
result in low performance of GML application and destroy 
abundant semantic of GML data. Thus, appropriate GML 
storage approach would improve extend and depth of GML 
application. Although GML application increasingly manifold, 
current research on GML database and related technologies are 
on the germination status. Neither clear concepts nor 
architecture nor mature special product exists. 
Earlier XML/GML data is stored as document (text file), is 
queried by information access methods such as key word query 
which is simple and fit for frequently document updating 
context. Since such storage approach lacks of systematic 
storage and query mechanism which leads to low query ability 
and can neither satisfy complex condition query nor optimize 
query. 
Traditional relational database system is more mature on 
aspects of query, concurrency, security and other technologies, 
and extends function of processing XML/GML data. Due to the 
big difference between semi-structural XML data model and 
traditional relational model would bring problems on data 
storage and query. The problems are as follows: (Dwhen store 
XML/GML data, traditional RDBMS doesn’t support hierarchy 
and semi-structural data format, nested XML/GML data should 
be converted by mapping mechanism before stored in simple 
relational tables. Such approach not only reduces storage 
efficiency but also leads to information losing during data 
conversion; ©when query XML/GML data, XML/GML query 
requirement should be firstly converted to database query 
expression, such as SQL; Then query engine of database 
optimize query expression and generate query execution plan; 
finally convert query result to XML/GML data. Such approach 
resolves problems of query complexity in a certain extent, but 
too many conversions bring new problems of low efficiency 
and confusion of query semantic. 
As increasingly mature of XML technology, XML databases 
have been broadly studied and applied, databases used for 
storing and managing XML data have come forth, namely 
native XML database [www.rpbourret.com, 2005]. Advantages 
of native XML databases are as follows[Liugang,2005]: 
©effectively support XML character of self-description, semi 
structure and sequence; ©provide “out and home access” for 
XML document, system store XML data directly, but not 
convert XML data to relational model or object orientated 
model, and store XML document in relational database or 
object orientated database. The native XML database can store 
XML document in native XML database and has capability to 
access the “same” document; ©directly support XML query 
Corresponding author.